Structural properties are well reproduced by all models (Table 2), but the significant improvement of our potential stands in the elastic constants which relate to how the system responds to stress. Indeed, structure and elasticity are important parameters for elucidating grain boundary stability. All potential models correctly predict the relative stability of the defect energies. The Morelon potential model performed best as it was specifically derived to replicate defect formation energies, but it largely underestimates the bulk modulus. The energies calculated with the Morl and the Arima potential models are overestimated; this is a known disadvantage of using rigid ion models as the ionic polarisability is not taken into account. For completeness, we report two shell models with the best results given by the Catlow potential model. The Morl, along with the Grimes shell potential model, accurately reproduce the activation energy of oxygen migration (the migration path was the lowest energy and most favourable diffusion mechanism observed in bulk UO2 [1]). The major deficiency of the Morl potential is that the cation defect energies are high, and hence the number of cation defects will be underestimated. However, this should not be an issue unless this model was applied to processes such as grain growth where cation mobility will contribute.
